AI in Focus - Learning series
Watch us build with AI - live.
Every week, thoughtbot founder Chad Pytel codes with AI in real time, exploring AI coding, how LLMs integrate into Rails applications and what it takes to turn AI ideas into production-ready software.
About this series
What is AI in Focus?
AI in Focus is our long-running weekly livestream dedicated to building real AI software. We stream to YouTube, Linkedin, and Twitch.
Since 2024, Chad and his rotating guest co-hosts have been coding live with generative AI tools, exploring how large language models (LLMs) integrate into Rails applications and modern product stacks. Each episode goes beyond theory. We test ideas in real time, examine tradeoffs, and discuss the practical constraints of shipping AI-powered features.
We cover implementation patterns, cost considerations, data privacy, ethics, feasibility and always stay grounded in the realities of building and maintaining quality software.
AI in Focus is designed for product leaders, founders, and engineering teams who want to move from AI curiosity to production-ready systems.
What we're building now
Coding with Claude Code
Many of our recent episodes focus on coding with Claude Code: testing AI-assisted refactoring, evaluating code audits, experimenting with prompt-driven workflows, and pushing the boundaries of what LLMs can support in day-to-day Rails development.
We don’t just talk about what’s possible. We open the editor, write code, and examine what holds up in production and what doesn’t.
What's new
Claude Code x Rails
A new Claude Skill for Rails code audits
Chad tests out our new Claude Code skill that audits real Rails applications using thoughtbot’s patterns and best practices.
Part 1: Extreme rapid prototyping with Claude Code
Kevin demos how he’s building an agent with Claude to digest research, spike a landing page, then apply design-driven prompting to refine towards an end product.
Part 2: The dev side of Claude Code-assisted rapid prototyping
Louis dives into how he developed quality, production-ready code with Claude during a 2-week sprint for a new app.
Practicing TDD with Claude
The hardest part of AI-assisted TDD is keeping tests meaningful, not just passing. See how tests provide essential guardrails and why experienced humans remain responsible for quality.
Security and compliance
Working with AI in healthcare
Claude skills for FDA-regulated Health Tech
Chad and Damian explore how Claude Code skills can help draft and structure critical SaMD documentation for FDA compliance.
The new top_secret gem for redacting sensitive data
Steve demos our new gem for automatically removing personal names, emails, phone numbers, and other sensitive information from text before sending it to an external LLM.
Just getting started?
Bring your best AI idea to life
How do we separate valuable real-world applications from misleading AI market hype? Part 1 of our series featured thoughtbot product leaders sharing their experiences and advice for turning AI into a powerful tool in your problem-solving toolbox.
Whether you're stuck or just starting on your AI initiative, you'll gain actionable insights on everything from technology choices to managing your AI project after launch.
Part 1
Strategic Foundations
Getting started
The group discusses the most common misconceptions that hold teams back from starting AI projects.
The AI feasibility framework
Is AI right for your product? Learn how to evaluate your technical options, business impact, cost, and role in your product roadmap.
Managing AI after launch
The group discusses why the human element is critical to managing and iterating on your AI project for long-term success.
